Financial Firm Bankruptcy and Systemic Risk
Jean Helwege*
April 2009
*Department of Finance, Smeal College of Business, Penn State University, 318 Business
Building, University Park, PA 16802. Email: [email protected].
I am grateful for helpful comments from Brent Ambrose, Adam Ashcraft, Mark Carey, Lee
Crabbe, Ingo Fender, Ann Gabor, Jacob Gyntelberg, Beverly Hirtle, Jay Huang, Frank Keane,
Srinivasan Krishnamurthy, Paul Kupiec, Frank Packer, Stavros Peristiani, Lukas Roth, Eric
Santor, Nikola Tarashev, Andrew Zhang, Gaiyan Zhang and participants at the FDIC’s 19th
Annual Derivatives conference.
Financial Firm Bankruptcy and Systemic Risk
Financial firm distress often leads to regulatory intervention, along the lines of “too big to
fail” (TBTF) policies. Despite the moral hazard implications of such actions, regulators often
fear that the collapse of a financial firm will lead to negative spillovers into the real side of the
economy. The mechanism by which financial crises connect these firms to the economy is not
well understood, but two oft-cited channels are domino effects (counterparty risk) and the effects
of fire sales on markets, which may transmit to the economy through a financial accelerator. In
this paper we analyze the policy responses for avoiding systemic risk while considering the role
of these two factors. Evidence from past bankruptcies suggests that cascades caused by
counterparty risk do not occur, as most firms diversify their exposures. Instead, crises tend to be
symptomatic of common factors in financial firms’ portfolios, which lead to widespread
instances of declining asset values. These low asset values are often misinterpreted as resulting
from fire sales by distressed firms, but are more likely to represent massive losses from poor
investment choices. Fire sales in the U.S. are less likely because bankruptcy laws allow
substantial time for the disposal of impaired assets. In sum, the two main reasons for TBTF
policy, counterparty risk and fire sales, are unlikely to pose major risks to the system should
failed financial firms be permitted to file for bankruptcy protection. Instead, regulators focus
should be on broad approaches, such as fiscal stimuli, that will mitigate wealth effects for a large
group of consumers who have all suffered from exposure to common risk factors in their
investment portfolios.
Key words: Financial institutions, systemic risk, Too big to fail, fire sales, counterparty risk
JEL codes: G28, G33, E44, E58, E61
2 Systemic risk is the risk that the financial system will fail to function properly because of
widespread distress. Failure of the system implies that capital will not be properly allocated:
That is, positive net present value projects will not be undertaken because the collapse of the
system leads to a shortage of credit, possibly due to an inability or unwillingness on the part of
investors to fund risky projects that they might have pursued prior to the crisis. Such pervasive
financial fragility may occur because the failure of one firm leads to the failure of other firms
which cascades through the system (e.g., Davis and Lo (1999) and Jarrow and Yu (2001)). Or
systemic risk may wreak havoc when a number of financial firms fail simultaneously, as in the
Great Depression when more than 9,000 banks failed.1 The conventional wisdom is that
systemic risk should be prevented, or at least mitigated, in order to avoid the risk of a deep
economic downturn.
Much of the evidence supporting the need for strong regulatory action vis-a-vis systemic
risk comes from the Great Depression, which provides overwhelming proof that failure to
prevent systemic risk episodes can be extraordinarily costly to the real economy. Friedman and
Schwartz (1971) argue that the Great Depression would likely never have been as deep and
painful as it was had the Federal Reserve (the Fed) increased the money supply in the face of a
declining money multiplier (owing to a reduction in the amount of money passing through the
banking system as thousands of banks failed). Bernanke (1983) further posits that the financial
firm crisis of the Great Depression exacerbated the money supply problems with severe
disruptions in the credit allocation process. Japan’s Lost Decade provides further evidence that a
weak financial system can lead to disastrous effects on the real side of the economy. Despite the
fact that depositors in Japan were highly unlikely to lose any money from bank failures, and
1
See Benston (1986).
3 despite the tacit willingness of Japanese regulators to let economically insolvent banks continue
to operate, the weak banking system sharply raised the cost of uninsured funding for Japanese
banks (see Covrig, Low and Melvin (2004)) and simultaneously reduced the availability of new
bank credit.
Such extreme real effects that appear to be the result of financial firm fragility have led to
a large emphasis on the prevention of systemic risk problems by regulators. Foremost among
these policies is “too big to fail” (TBTF), the logic of which is that the failure of a large bank or
financial institution will have ramifications for other financial institutions and therefore the risk
to the economy would be enormous. TBTF was behind the Fed’s decisions to orchestrate the
merger of Bear Stearns (henceforth, “Bear”) and J.P. Morgan Chase (henceforth “JP”) in 2008,
its leadership in the restructuring of bank loans owed by Long Term Capital Management
(LTCM), and its decision in Fall 2008 to prop up the insurer American International Group
(AIG). TBTF may be justified if the outcome is prevention of a major downswing in the
economy. However, if the systemic risks in these episodes have been exaggerated or the salutary
effects of these actions overestimated, then the cost to the efficiency of the capital allocation
system may far outweigh any potential benefits from attempting to avoid another Great
Depression.
Considering that much of the fear of systemic risk appears to be the fear of creating
another Great Depression, we should consider how the financial system in the U.S. today might
respond to similar crises if the Fed were to let the events play out without interceding. Many
differences exist between that episode and any crises that might occur in the U.S. today. First,
the Great Depression occurred at a time before deposit insurance so the real decline in consumer
4 wealth as a result of bank insolvencies was substantial for a large number of ordinary Americans.
Moreover, even such large shocks to the system might have been relatively minor (i.e., a
recession instead of a deep depression) had the Fed followed a more expansive monetary policy
in the face of these failures. Given the Fed’s recent track record, it is inconceivable that a
multitude of bank failures could again be accompanied by such restrictive monetary policy.
Considering these differences, one could easily argue that regulator’s interventions in recent
financial crises are misguided.
No doubt, no regulator wants to take the chance of standing down while watching over
another systemic risk crisis, so we do not have the ability to examine empirically what happens
to the economy when regulators back off. There are very few instances in the modern history of
the U.S. where regulators allowed the bankruptcy of a major financial firm. Most recently, we
can point to the bankruptcy of Lehman, which the Fed pointedly allowed to file for bankruptcy.
However, with only one obvious case where TBTF policies were overturned, we have only an
inkling of how abandoning TBTF policy might affect systemic risk. Moreover, at the same time
that Lehman failed, the Fed chose to bail out other financial firms – money market mutual funds
(MMMFs), commercial paper issuers like GE Capital, and AIG. These bailouts had followed
only six months after the Fed arranged for the merger of Bear and JP, so expectations of a bailout
were running high at the time. All in all, observing one bankruptcy in the midst of several TBTF
bailouts makes it very difficult to draw conclusions about empirical patterns in the data.
Furthermore, at the time of the AIG bailout the Federal Reserve and the Treasury were
campaigning for Congress to allocate TARP funds and engaged in scaremongering about the
prospects of a second Great Depression to make the passage of TARP more likely. We will
never know if the market’s reaction following the Lehman bankruptcy reflected fear of contagion
5 from Lehman to the real economy or fear of the depths of existing problems in the real economy
that were highlighted so dramatically by Mssrs. Bernanke and Paulson.
Besides Lehman, a few other large financial firms such as Drexel and Finova, have also
been allowed to fail, but one can argue that they were not large enough to deserve TBTF status
or that they involved fraud, making it much less likely that investors would extrapolate the
experiences with those firms to the rest of the financial services industry. Nonetheless, these
cases and a few others in the sparse history of financial firm failures in post-Great Depression
U.S. allow us to analyze the mechanisms by which such risk could cause an economy-wide
collapse. We focus on two types of contagion that might lead to systemic risk problems: (1)
information contagion, where the information that one financial firm is troubled is associated
with negative shocks at other financial institutions largely because the firms share common risk
factors; or (2) counterparty contagion, where one important financial institution’s collapse leads
directly to troubles at other creditor firms whose troubles snowball and drive other firms into
distress. The efficacy of TBTF policies depends crucially on which of these two types of
systemic risk mechanisms dominates. Counterparty contagion may warrant intervention in
individual bank failures while information contagion likely does not.
If regulators do not step in to bail out an individual firm, the alternative is to let it go
bankrupt. In the case of a failed commercial bank or other depository institution, the process
involves the FDIC as receiver and the insured liabilities of the firm are very quickly repaid. For
example, in the recent failure of IndyMac Bank, the FDIC allowed depositors to withdraw their
insured funds through automatic teller machines, suggesting minimal disruptions for these
depositors. Other creditors may or may not receive partial repayment of their funds in a timely
6 fashion (see Flannery and Sorescu (1996) and DeYoung and Reidhill (2008)). In contrast, the
failure of an investment bank or hedge fund does not involve the FDIC and may closely resemble
a Chapter 11 or Chapter 7 filing of a nonfinancial firm. However, if the nonbank financial firm
in question has liabilities that are covered by the Securities Industry Protection Corporation
(SIPC), the firm is required by law under the Securities Industry Protection Act (SIPA) to
liquidate under Chapter 7.2 This explains in large part why only the holding company of
Lehman filed for bankruptcy in 2008 and its broker-dealer subsidiaries were not part of the
Chapter 11 filing.
A major fear of a financial firm liquidation, whether done through the FDIC or as
required by SIPA, is that fire sales will depress recoveries for the creditors of the failed financial
firm and that these fire sales will have ramifications for other firms in related businesses, even if
these businesses do not have direct ties to the failed firm (Shleifer and Vishny (1992)). This fear
was behind the Fed’s decision to extend liquidity to primary dealers in March 2008 - Fed
Chairman Bernanke explained in a speech on financial system stability that
“the risk developed that liquidity pressures might force dealers to sell assets into
already illiquid markets. This might have resulted in …[a] fire sale scenario…, in
which a cascade of failures and liquidations sharply depresses asset prices, with
adverse financial and economic implications.”3
The fear of potential for fire sales is expressed in further detail in the same speech as a reason for
the merger of Bear and JP. The idea is that without the merger, Bear would have been forced
into bankruptcy and the dire consequences of fire sales would have ensued:
2
See Don and Wang (1991). Bernanke’s May 13, 2008 speech at the Federal Reserve Bank of Atlanta financial conference at Sea Island,
Georgia. The identical speech was presented again to the Risk Transfer Mechanisms and Financial Stability
Workshop, Basel, Switzerland (the BIS) on May 29, 2008.
3
7 “Bear advised the Federal Reserve…that it would be forced to file for bankruptcy
the next day unless alternative sources of funds became available. A bankruptcy
filing would have forced Bear's secured creditors and counterparties to liquidate
the underlying collateral and, given the illiquidity of markets, those creditors and
counterparties might well have sustained losses. If they responded to losses or the
unexpected illiquidity of their holdings by pulling back from providing secured
financing to other firms, a much broader liquidity crisis would have ensued….the
Federal Reserve Board judged that it was appropriate to use its emergency lending
authorities …to avoid a disorderly closure of Bear.”
Bernanke’s description of the bankruptcy process highlights the idea that creditors of a failed
firm are forced to liquidate assets, and to do so with haste. However, U.S. bankruptcy laws are
set up with the idea that creditors should be allowed to maximize the value of the assets now
under their control. If that value is maximized by continuing to operate, the laws allow such a
reorganization of the firm. Admittedly, there is a prejudice against reorganization of financial
firms, due to concerns about allowing fraudulent securities sales to continue unhampered, but
there is certainly no prejudice against selling assets in an orderly procedure to maximize the
value to creditors, be they secured or unsecured.
Bankruptcy actually reduces the likelihood of fire sales because assets are not sold
quickly once a bankruptcy filing occurs. Upon filing for bankruptcy, cash does not leave the
firm without the approval of a judge. Without pressure to pay debts, the firm can remain in
bankruptcy for months as it tries to decide on the best course of action. Indeed, a major
complaint about the U.S. code is that firms can easily delay reorganizing and slow down the
process (Bradley and Rosenzweig (1992), Weiss and Wruck (1998), Bebchuk and Chang (1992),
Wruck (1990) and Weiss (1990)). If, however, creditors and management believe that speedy
assets sales are in their best interest, then they can press the bankruptcy judge to approve quick
action. This occurred in the case of Lehman’s assets involving workers, who might have given
up on the firm and left haphazardly were their divisions not sold quickly with the judge’s
8 approval. Concerns that a bankruptcy filing will force the firm to engage in fire sales are
unwarranted given that bankruptcy laws are set up to allow time for an orderly process of dealing
with the distressed firm’s assets.
Not all countries have the same bankruptcy rules, and even within the U.S., the rules
differ for financial firms with SIPC or FDIC coverage. Bankruptcy may be viewed as a harsher
outcome for some firms compared to others. In particular, a tendency toward liquidation that
destroys value may drive firms to restructure out of court to reduce this inefficiency (Haugen and
Senbet (1978)). If so, then fire sales might occur before the firm files for bankruptcy as a last
chance attempt to stave off liquidation. Nonetheless, the extent to which fire sales occur outside
of bankruptcy is a function of the stringency of the bankruptcy code: The more lenient the code
toward debtors, the less likely a firm will find it optimal to engage in a fire sale out of court to
avoid bankruptcy. In the U.S., bankruptcy cases that are not automatically receiverships are not
harsh in the sense that debtors have time to consider their alternatives (Smith and Stromberg
(2004)). In the bankruptcy court supervised sale of a fraction of Lehman assets to Barclays
Global, the pace was extraordinarily quick (although considerably slower than the sale of Bear to
JP) but the sale only involved assets that were considered likely to deteriorate in value if not sold
quickly (i.e., assets involving employees). The remaining assets are likely to be sold off in a
much slower fashion as investors struggle to determine the true value of credit losses. As a
result, neither creditors nor shareholders are taken advantage of by bottom-fishing merger
partners.
In order to see the potential benefits of various regulatory responses to a financial crisis,
including that of bankruptcy, we must first consider the causes of financial firm distress. These
9 are discussed in section 2. The cost of bankruptcy and other responses that aim at minimizing
moral hazard issues depend on the extent to which formal bankruptcy risks systemic effects on
the economy. While we cannot know with certainty the potential for an outcome such as the
Great Depression, we can examine the impacts of previous financial firm failures. This is the
topic of section 3, which discusses the cases of Finova, Drexel and the Bank of New England.
Section 4 considers various regulatory responses to the distress of financial firms. Section 5
concludes.
II. The Causes of Financial Crises
The appropriate policy response to a large financial firm failure depends crucially on the
nature of the collapse. Society might benefit if regulators salvage assets in the failed firm and
keep the entity as a going concern, regardless of how the firm value landed at so low a level.
However, one must also consider the signal sent to investors if firms receive aid despite having
made poor choices that caused the drop in firm value. Essentially, the appropriate policy
response weighs the trade-off between preventing systemic risk and preventing an atmosphere of
reliance on government handouts (the so-called moral hazard problem). That trade-off depends
critically on the effectiveness of the measures intended to prevent systemic risk, which in turn
depends on the nature of the crisis: counterparty risk or information contagion. In the former
case, the collapse of one bank leads directly to the collapse of another and so forth (e.g., Jarrow
and Yu (1999), Davis and Lo (1999), Giesecke (2002)). For example, if we observe ten failed
banks that suffered only from counterparty risk, we can draw a line connecting each one’s
liabilities to the failure of another in the set of ten. With information effects, there is no such
line. Instead, all ten banks share a set of risks in their portfolios. Even if none is a creditor in the
10 other’s bankruptcy, all ten will experience distress simultaneously (or at least within the same
short time period), because all ten invested in the same types of assets which now have
substantially lower values. We refer to this as information contagion because frequently the ten
banks do not fail exactly simultaneously. Rather, one bank fails first and investors in the other
nine banks correctly infer that their banks are also in trouble and act accordingly.
Rochet and Tirole (1996) argue that regulators can intervene in two ways: they can help
out the bankrupt bank or they can help out the creditors of the bankrupt bank. Implicitly, they
are assuming counterparty is the main source of financial firm distress. If regulators help out the
creditors, rather than the initial bank, the initial distressed bank is allowed to go under, imposing
a penalty on the financial firm for failing to control the credit risk of its portfolio. This sends a
signal to investors to carefully monitor their assets and choose investments wisely. At the same
time, the choice to help out the creditors prevents a widespread crisis. This is because the
regulators would also involve inject capital into creditor firms of the failed bank if they are at
risk of failing themselves as a result of the bankruptcy.
A. Counterparty Contagion
In a case of counterparty contagion, the initial bank gets into trouble and the other banks
become distressed as a result (e.g., Davis and Lo (1999)). Commercial banks make loans to each
other (Rochet and Tirole (1996)) and investment banks make loans to and/or invest in other
financial firms. Thus, the financial sector is at risk for systemic problems because many firms
have exposure to the failed firm. Implicitly, this scenario assumes that whatever caused the first
bank to lose its capital did not cause other firms to lose their capital at the same time. For
11 example, in the case of Bear, Fed Chairman Bernanke described the situation to Congress4 as
“extremely complex and interconnected.” He further declared:
“Bear Stearns participated extensively in a range of critical markets. The sudden
failure of Bear Stearns likely would have led to a chaotic unwinding of positions
in those markets and could have severely shaken confidence. The company's
failure could also have cast doubt on the financial positions of some of Bear
Stearns' thousands of counterparties and perhaps of companies with similar
businesses. Given the exceptional pressures on the global economy and financial
system, the damage caused by a default by Bear Stearns could have been severe
and extremely difficult to contain. Moreover, the adverse impact of a default
would not have been confined to the financial system but would have been felt
broadly in the real economy through its effects on asset values and credit
availability. To prevent a disorderly failure of Bear Stearns and the unpredictable
but likely severe consequences for market functioning and the broader economy,
the Federal Reserve, in close consultation with the Treasury Department, agreed
to provide funding to Bear Stearns through JPMorgan Chase.”
Similarly, in testifying to Congress5 about the bailout of AIG, Bernanke implied that the firm’s
counterparties may have defaulted had the insurer been allowed to fail:
“....it was well known in the market that many major financial institutions had
large exposures to AIG. Its failure would likely have led financial market
participants to pull back even more from commercial and investment banks, and
those institutions perceived as weaker would have faced escalating pressure.
Recall that these events took place before the passage of the Emergency
Economic Stabilization Act, which provided funds that the Treasury used to help
stem a global banking panic in October. Consequently, it is unlikely that the
failure of additional major firms could have been prevented in the wake of the
failure of AIG. At best, the consequences of AIG's failure would have been a
significant intensification of an already severe financial crisis and a further
worsening of global economic conditions. Conceivably, its failure could have
resulted in a 1930s-style global financial and economic meltdown, with
catastrophic implications for production, income, and jobs.”
4
Testimony on April 3, 2008 to Committee on Banking, Housing, and Urban Affairs, U.S. Senate.
5
Testimony on March 24, 2009 to Committee on Financial Services, U.S. House of Representatives.
12 In such cases of counterparty risk, the interconnectedness of the financial firms makes it highly
unlikely that a second financial firm could withstand losses induced by the first firm. While the
idea of a domino effect of one firm failing and starting a cascade of additional failures seems
eminently plausible, empirically such a failure of one bank could only cause the financial distress
of other banks and therefore a systemic crisis if a number of conditions were to hold. First, the
initial bank must be large (hence the TBTF policy). Second, it must have experienced a large
decline in the value of its assets (so the losses imposed on other firms are substantial). In
addition, for such a cascade to continue throughout the system, the losses imposed on these
creditors must be a large fraction of their assets, which is to say that the creditors of this large
bank must be quite undiversified. Otherwise, the losses at the first bank will not be great enough
to cause a crisis in the second bank. It is an empirical question as to whether this “perfect storm”
combination of events is likely to occur in nature and be a representative example of the cause of
financial crises.
In order to conclude that counterparty risk is a major cause of financial crises, we must be
able to conclude that the crisis originates at the initial bank. This suggests that the shock that
drove the initial bank into distress was idiosyncratic in nature – otherwise, if other banks also
suffered major shocks, it would not be appropriate to describe the initial bank as causing distress
in other firms. In the setting of Rochet and Tirole (1996), the problem at the initial bank is poor
credit analysis and monitoring. Had the initial bank done its job correctly, it would not be the
subject of discussion of a bailout to prevent systemic risk. Idiosyncratic shocks related to poor
banking practices are often related to fraud or at least a business plan that was especially
reckless, perhaps because of the overzealous or pernicious actions of one person or people at one
13 firm. One can point to Drexel, Barings and Franklin National Bank as instances where criminal
activity was charged and the public viewed it as the dominant reason for the firm’s collapse. Not
surprisingly, the policy responses in these cases were not cases of TBTF – the desire to punish
poor investment choices far outweighed concerns over the potential loss of confidence in the
system or counterparty risk. Nonetheless these failures did not lead to systemic risk. While it is
possible these firms did not have much impact on the system as a whole because they were too
small to qualify as TBTF firms, the fact that they had little in the way of ripple effects casts
serious doubts on the potential for counterparty risk and the cascade mechanism by which it
affects the real economy.
If the cause of distress at the first institution is idiosyncratic, the subsequent losses are
less likely to be very large compared to a scenario where many assets systematically lose value.
For counterparty risk to be a major concern, idiosyncratic losses at the initial bank would have to
be widespread losses on the portfolio that stem from a credit policy that differs sharply from
those of other financial institutions. Moreover, in a setting such as that of Rochet and Tirole’s
(1996) and David and Lo (1999) and other contagion models, the second, third and other banks
only follow into distress because they have ties to the initial bank. If a bank is not a creditor of
the initial bank, it will not require assistance. Systemic risk then requires strong ties to the initial
bank and strong ties among the remaining banks.
Whether or not this situation is of major concern to regulators is an empirical question,
and, unfortunately, one for which data are quite limited. Nonetheless, the empirical evidence to
date suggests that no such domino effect would take place were regulators to abandon TBTF
policies. Of course, the situation could be more extreme for the very largest financial institutions
14 that have not been allowed to fail, but studies of bankrupt companies to date point to very limited
counterparty effects on other firms. Lang and Stulz (1992), Theocharides (2008) and Jorion and
Zhang (2007) study the effects on stock prices, bonds, and credit default swap (CDS) rates,
respectively, of other firms in the industry when a bankruptcy occurs. If counterparty effects on
suppliers are very large, the other securities in the industry should experience a severe drop in
value upon the announcement of the initial bankrupt firm. While these studies find statistically
significant effects, the relatively small impact suggests that suppliers in the bankrupt firms’
industries do not collapse in a cascading fashion. Hertzel, Li, Officer and Rodgers (2008)
examine suppliers and customers of the distressed firms that list the bankrupt firms in their 10Ks. The SEC requires publicly traded firms to list such financial relationships if sales to one
firm account for at least 10 percent of the total. Hertzel et al. find that customers are not affected
by the filing but that suppliers are. However, the average effect for suppliers is less than 2% of
the market value of equity, hardly enough to force the average supplier into distress.
A study that is more specifically focused on the issue of counterparty risk is that of Jorion
and Zhang (2008), henceforth JZ, who show that the effects on counterparties are remarkably
small.6 Their data cover bankruptcies that were filed during1999-2005 and include the claims
owed to the top twenty unsecured creditors in each filing.7 Because these were Chapter 11
filings, most of the bankrupt firms are nonfinancial entities, with only a small fraction of the
6
Kabir and Hassan (2005) also focus specifically on counterparties in a case study of the failure of LTCM. They
also find small impacts, as the decline in the market value of the equity of LTCM’s 14 counterparties was only 6%,
again hardly enough to cause a second bankruptcy.
7
Secured creditors would obviously experience even smaller counterparty losses given the collateral backing their
loans. 15 cases involving financial firms.8 However, one would expect the effects for nonfinancial firms
to be stronger than for financial firms, as their suppliers are not required by regulators to have a
diversified cash flow. For example, it would not be surprising to find that the collapse of an auto
manufacturer leads to the collapse of an auto parts supplier because of firm specific investments
made by the supplier. We expect the supplier to sells its auto parts to only one automaker. We
also expect that the parts supplier will be fairly concentrated in its focus (i.e, to be small and only
have one segment), so it will have few other sources of cash flow besides the sales of auto parts.
Thus, the extremely undiversified asset base of this nonfinancial firm would be very susceptible
to the collapse of its major customer. In contrast, we would not expect financial firms and their
creditors, who are also financial firms, to involve such large exposures.
Financial firms rarely have such focused business plans, as they actively seek to diversify
their assets. For example, a bank that lent money to the same auto manufacturer would have
much less exposure than our hypothesized parts supplier because bank regulations limit the size
of any one loan. Specifically, loans to one borrower may not exceed fifteen percent of the bank’s
capital. And few banks would lend to the point where they find that constraint binding, as the
profitability of such a loan would be more than offset by its potential for causing bankruptcy.
Instead, large loans that would unduly concentrate a bank’s portfolio are typically underwritten
by a syndicate. By the same logic, if a bank or insurance company or broker were to fail, it is
unlikely that their top twenty creditors would have a very large portion of its assets tied up in this
8
Many financial firms are not eligible to file for Chapter 11 because they have regulated businesses that would by
law be required to liquidate (Chapter 11 is not intended for liquidations but reorganizations). If a stock broker goes
bankrupt, its accounts will fall under the regulation of SIPC and SIPC will oversee its liquidation. Financial firms
whose liabilities include deposits would also not enter Chapter 11, as bank regulators would put the entity into
receivership. Likewise, state insurance commissioners would take control of insurers. Financial holding companies
who own equity in bank, insurance or brokerage subsidiaries could enter Chapter 11 if they do so without their
subsidiaries (e.g., Lehman in 2008). 16 firm. Furfine (2003) shows that counterparty exposures in the interbank loan market (the fed
funds market) is quite small, and therefore are probably not large enough to present a systemwide threat to the banking system.
Anecdotally, we can point to the bankruptcy of Lehman in September 2008, which filed
at the holding company level with $600 billion in liabilities. The largest single creditor reported
in the bankruptcy filing was Aozora Bank, based in Tokyo, which was owed a mere $463
million, or less than one tenth of one percent of the claims in Lehman’s filing.9 While a figure
that is only a tad under a half billion is not a trivial loss, Aozora made clear that it had offset
much of the exposure, estimating its loss at only $25 million (in part because it had purchased
insurance on its Lehman claim with a credit default swap). Thus, the expected loss of the largest
creditor is quite small, especially when compared to the $7.4 billion capital base of Aozora.
The Lehman Aozora anecdote appears to be quite representative of the general situation,
given the evidence in JZ’s sample of 694 creditor-events. In their sample, a financial firm,
Citibank, is owed the largest amount of any claim in the database ($1.75 billion). Nonetheless,
as with Aozora, that claim is small relative to the size of Citibank’s portfolio. We can see from
the exposure statistics in their Table V, the largest percentage exposure among the 111 claims
owed to financial institutions is only 2.39 percent of market value. As a percentage of assets,
this figure is many times smaller because banks are highly levered entities. Moreover, in the
case of Citibank’s $1.75 billion claim, the bankrupt firm, Enron, did not lose so much value that
9
Bankruptcy regulations require the failed institution to list the top 20 unsecured creditors (presumably the creditors
with the most to lose, as secured creditors have collateral to back their debts). Often, the largest creditor on the list
is a trustee for unsecured bondholders. The trustee is not at risk of losing money, only his clients. As the clients are
not listed separately it is difficult to determine who among the unsecured creditors is truly the single largest.
However, as bankruptcies proceed the holdings of the bonds typically move into the hands of vultures, who rarely
own more than a third of any one bond class, suggesting the original list of bond creditors had much lower holdings.
17 Citibank had no recovery on the loan: they only wrote down the value by $228 million, so their
loss was only a fraction of the claim filed in the bankruptcy court. Most claims are substantially
smaller, measured either by dollars or percent: the mean claim in the sample is $11.5 million
and the mean exposure as a percent of market value is 16 basis points. Many of these are trade
credits, which are very small. However, even restricting the claims to those owed to financial
sector specialists, bonds and loans, the average claims are $17.1 million and $163.6 million,
respectively. Recall that these claims are among those owed to the top 20 unsecured creditors, so
the average claim in the bankruptcy is much smaller.
More often, the largest claims in percentage terms arise with suppliers and not financial
institutions. JZ investigate all creditors in the sample who have one or more customers that
account for at least 10 percent of their revenue by examining 10-K filings. None of these firms
with such strong reliance on one customer are financial firms. But even among the nonfinancial
firms, the incidence of extreme concentration of sales is rare. Only six firms in the sample list
the bankrupt firm as a major customer in their 10-Ks and the average exposure among these
firms is only 11 percent.
JZ’s sample supports our conjecture that most firms have diversified cash flows or at
least are diversified across customers. Few firms suffer dramatically from the bankruptcy of a
single firm with which they have business ties because few firms have such strong ties that a
bankruptcy would throw them into distress. Not surprisingly, then, do we find that the 251
bankruptcies in JZ’s sample rarely trigger a cascade of subsequent bankruptcies: only 22
creditors in the sample delist within two years of the 251 initial bankruptcies. Furthermore,
inferring that 22 cases of distress occurred as a result of the 251 bankruptcies overstates the
18 incidence of counterparty-induced bankruptcy because not all of these delistings can be tied to
the original 251 failures. Moreover, the problems at these 22 creditor firms can hardly be
described as part of a cascade of failures, as no other firms failed as a result of problems at the 22
affected firms. A cascade set off by counterparty risk is a situation where one firm’s failure
causes another firm’s failure which causes a third firm’s failure on so on. If counterparty risk
were to explain the event of ten firms failing in a short period of time, we would have to see that
the first firm causes the second firm’s problems, which in turn causes the third firm’s distress,
and so on up to ten. JZ’s data indicate that there is never a third firm involved, let alone a fourth,
fifth, sixth, seventh, eighth, ninth or tenth.
What if the evidence in JZ’s sample is unrepresentative of what would happen in a
serious disaster? After all, their sample includes seven years of which only two are recession
years and the 2000-2001 recession was not considered especially severe. Perhaps the incidence
of second and third defaults is understated in their data because they do not examine a period as
disastrous as the Great Depression. Again the data are not available for such a study, but we can
consider analytically the difference in bankruptcies that occur because of systematic risk and
those that arise from idiosyncratic risk. Recall that the important issues are the severity of the
shock and the degree of diversification of the counterparties. We have no reason to believe that
counterparties would be more or less diversified in different economic settings, but we do know
that the severity of the shock to the first firm is likely to be greater when the source is
idiosyncratic. Idiosyncratic risk is substantially larger than systematic risk (marketwide
volatility or industry volatility) for the typical listed firm (see Campbell, Lettau, Malkiel and Xu
(2001)). A priori, then, the first firm’s shock on average would be lower in times when the
19 shocks are macroeconomic in nature and therefore would decrease the potential for failure by a
second firm relative to the events studied by JZ.
B. Information Contagion
Rather than counterparty risk, financial crises are more often situations where we observe
dozens of distressed firms in the same time period because the firms have a common factor
causing simultaneous incidences of financial distress. For example, in the credit crisis starting in
2007, we observed major losses and/or failure at Merrill Lynch, UBS, Bear Stearns, Lehman,
Washington Mutual, Countrywide, AIG, Fannie Mae, Freddie Mac, and Indy Mac. These ten
firms rarely had strong business ties to each other- Lehman’s bankruptcy filing did not list any of
these troubled firms as major creditors. Rather, the ten firms share a common exposure to the
mortgage market, specifically the subprime mortgage market. Thus, when Bear failed, the major
effect on the market was to trigger a re-evaluation of other financial firm’s health. Not
surprisingly, once Bear got into trouble investors focused considerable attention on Lehman, the
investment banking firm that most closely resembled Bear.
In times of financial upheaval, investors look into their portfolio with greater scrutiny and
try to identify securities that have something in common with those of the first failed firm. The
common factors in these episodes may be related to industry, such as the collapse of oil prices
leading to widespread defaults by independent oil companies in the mid-1980s (see Helwege and
Kleiman (1990)) or the extreme exposure to commercial real estate in the insurance industry
around 1990 (see Fenn and Cole (1994)). Or, it may be a common problem regarding the degree
to which investors understand the risk of the firm, either because of fraud or imperfect
accounting information as in Enron or Worldcom (see Duffee and Lando (2001), Collin20 Dufresne, Goldstein and Helwege (2004) and Yu (2005)). Or it may be a common factor related
to a macroeconomic variable, such as the failures of S&Ls in the early 1980s due to interest rate
risk (Helwege (1996)).
Each firm is exposed to a particular factor, and that factor moves in such a way as to
cause a negative shock to security prices. One can think of these shocks as normally distributed,
or at least drawn from a distribution that has more mass at the mean than at the extremes.
Among all the financial firms that experience a negative shock associated with this factor, we
expect that the ones with the most negative shocks will be among the first to become insolvent.
Thus, when we observe ten firms enter distress because of an event like the subprime crisis of
2007, a reasonable conclusion is that the first firm to catch the attention of regulators is the one
in the worst situation. Investors in the other nine firms watch how this firm is treated; they glean
new information from the announcements regarding the problems as well as from the ways in
which the regulators attempt to solve the problem; and they trade in the marketplace on this new
information, leading to updated valuations of the securities associated with the other nine firms.
The other nine firms’ stocks are likely to fall as a result of this re-evaluation – it is highly
unlikely that the first firm will attract the attention of regulators and those regulators will
announce that the firm is fine and their opinion will carry weight for some length of time (the
SEC did make such a statement regarding Bear but within a few days the Fed said they were on
the verge of bankruptcy and likely to cause a financial meltdown of the entire system). As the
market value of these firms approaches zero, they may seek capital to bolster their positions.
That capital, if successfully raised, will be at a very low price.
21 Empirical studies of past failures suggest information effects are extremely important.
Lang and Stulz (1992) find that stock market reactions to nonfinancial firm bankruptcies can be
positive, reflecting the fact that a bankruptcy filing of one firm may give investors confidence
that excess capacity in the industry will finally be taken off line and that the firm’s competitors
will gain market share going forward. Of course, not surprisingly, the information effects are
often negative. Lang and Stulz also find that some bankruptcies lead to very negative stock
market effects for other firms in the industry, reflecting investors’ revaluations of assets in that
industry. Similar results are found in Jorion and Zhang (2007, 2008) and Theocharides (2008).
Collin-Dufresne et al. (2004) consider the impact of credit events that affect bondholders and
find strong reactions in the rest of the bond market, even when no such event directly concerned
the other bonds. Crabbe (1991) shows that credit events involving leverage buyouts (LBOs) led
bond investors to charge higher premia on new bonds that might subsequently undergo an LBO,
but which were currently unaffected by such leverage. Sovereign risk researchers have long
argued that emerging market countries with no obvious economic ties to a failing country are
affected by its crisis (e.g., Kamin and Von Kleist (1999), Eichengreen and Mody (2000)).
Insurance company failures in the early 1990s, investigated by Fenn and Cole (1994) and
DeAngelo, DeAngelo and Gilson (1994), suggest that investors revalued their insurance stocks to
reflect the dire news that First Executive Corporation was likely to be insolvent.
Regardless of the nature of the common factor that causes these various negative shocks,
in these situations it is quite ineffectual to apply TBTF policies to the first failed firm. Helping
out this firm, which has no ties to the other firms that are affected, does nothing to improve their
balance sheets. Indeed, one can argue that helping out the first firm brings extra attention to the
other firms who might have muddled through the crisis were they not under such pressure to
22 improve their capital positions. In any case, money pumped into the first firm is unlikely to find
its way to the other firms. Helping out the first defaulting firm will have no impact on the losses
incurred by the second, third and fourth firms to default if the firms have no business ties to each
other. In contrast, the government to spend the same aggregate amount and divvy it up among
all the affected firms. In the latter case, all the firms would be impacted by the aid and the ones
with smaller shocks would be more likely to survive as a result.
An important element of financial crisis analysis is the mechanism by which information
contagion leads to systemic risk. If many financial firms are exposed to a risk factor that
experiences a negative shock, a downturn in the financial sector of the economy may easily
follow. But how does this turn into real effects on GDP? Without a complete understanding of
the channel by which their troubles become the troubles of the entire economy, we cannot speak
to the issue of how regulators should respond. A good starting place for thinking of the
transmission of financial shocks to the real economy is Friedman and Schwartz (1976). They
argue that financial crises lead to shrinkage of the financial system. In the case of the Great
Depression, the shrinkage arose directly from the fact that bank depositors withdrew funds and
did not put them into other institutions that might circulate the money, so the net effect was less
money circulating through the financial system. Given that the financial system has a money
multiplier effect that amplifies high powered money into the broader money supply, a decrease
in deposits means a decrease in the money supply. Friedman and Schwartz argue that the
decrease in the money supply was not offset by the Fed. Without a concomitant reduction in
prices, MV=PQ implies that the quantity of goods produced in the economy will drop.
23 Bernanke (1983) argues that these effects are unlikely to be the whole story because the
magnitude of the decline in the real side is out of proportion to the decline in the money supply.
Instead, he claims that the financial system carnage relates to the real side through the cost of
intermediation – banks make money by borrowing and lending, but if the spread between them is
too high (reflecting the high costs of analyzing borrower’s credit risk) then the amount of lending
will fall and projects will go unfunded. In this case, the negative shocks to financial firms work
their way through the system because these firms pull back in their lending (as they find that
funding is costlier or that credit analysis takes more effort and time), which in turn leads to less
investment in the economy and weaker prospects for economic activity. Bernanke, Gertler and
Gilchrist (1999) hypothesize a financial accelerator mechanism, whereby distress in one sector of
the economy leads to more precarious balance sheets and higher credit spreads to reflect the
increased risk of default. This leads to a drop in investment, which in turns leads to less lending
and widespread effects throughout the economy, followed by even less bank lending.
Another possibility is that none of the affected firms suffered from fundamental weakness
in their business lines, but were instead the victims of short-sellers and other miscreants who
worked to destroy confidence in these financial firms. In this case, the mechanism from the
financial side of the economy to the real side is illusory – the crisis affects both aspects of the
economy simultaneously through the destruction of confidence. Lost confidence causes
consumers to put off big ticket items and to move their savings into very safe products, reducing
the amount of capital available for new projects. The drop in confidence also reduces forecasts
of business profitability, leading to fewer new business starts and reduced investment among
existing businesses. The net effect of all these pullbacks leads to a smaller economy. Such a
24 situation could occur with equal likelihood were the initial incidence of distress a large
nonfinancial firm such as General Motors, or a financial firm such as Bear.
Although short-sellers have an incentive to repeat hurtful rumors about a company and
destroy confidence in the firm, they do not work in a vacuum. At the same time that they are
encouraging pessimism, the management and the existing shareholders are trying to spin a story
of optimism. Management and equityholders have an incentive to pass off the company as
healthier than it really is because both parties want the firm to continue as a going concern.
Management is fearful of losing their jobs so they neither wish to allow the firm to cease
operations as an independent entity nor to own up to mistakes if such honesty makes it more
likely they would lose their jobs (Gilson (1989)). Equityholders want the firm to continue so that
their stock has a chance to come back to a reasonable level (Geske (1977)). If the firm files for
bankruptcy, the odds of a substantial payout to shareholders are quite low, but shareholders still
attempt to gain value at the expense of creditors by putting off the final day of reckoning, in
hopes that the franchise value will improve in the meantime (Wruck (1990)). This delaying
strategy is more effective for large nonfinancial firms in bankruptcy because the U.S. code
presumes reorganization will occur for these firms. In contrast, Chapter 11 cases for financial
firms are more often a form of liquidation, which means equityholders are far more likely to
have their claims cancelled in bankruptcy. In other words, bankruptcy is harsher for
equityholders and management if the firm’s business involves the management of financial
assets. As we noted before, harsher bankruptcy rules encourage more out of court workouts. In
the case of a bank, an investment bank or an insurance company, capital regulations require that
the firm show not just solvency but a level of solvency that exceeds a regulatory minimum. If
the firm can stave off the day when it acknowledges a deficit in this regard, equityholders and
25 management benefit from the option to delay. The obvious way to avoid recognizing a lack of
capital when credit losses arise is to obfuscate the true condition of the portfolio – if the portfolio
were fine, there would be no need to exaggerate its health, but for underwater assets admitting
the truth is tantamount to ending the option to continue. In sum, while short-sellers have a great
incentive to exaggerate the problems of the firms in hopes of driving down the stock price, the
incentives for management and shareholders are even greater to overstate the health of the firm
as it approaches financial distress.
Throughout much of 2008, short-sellers were blamed for driving down financial firm
stock prices and the SEC went so far as to outlaw short-selling of financial firms stocks for
several months. The logic of this action was that calling attention to risky firms is sufficient to
drive their stocks down to the point of financial distress. In hopes of cancelling out these forces,
the SEC essentially tried to eliminate the opportunity to express a negative view on a financial
firm. For such a strategy to make sense for the economy, the regulators must decide that some
firms have sound businesses but are under (undeserving) attack. Management and equityholders
will try to convince regulators that their firms deserve more time to regain their footing, but this
puts regulators in the unenviable position of identifying truly worthy firms at a time when all the
owners and management are trying their best to put a positive spin on the situation and the
market is sending the contrary message.
The case of Lehman in 2008 suggests that regulators are heavily influenced by these
claims of fundamental soundness. Immediately after the Bear/JP merger, investors reacted by
asking “Which firms look like Bear?” and attention immediately gravitated to Lehman. The two
investment banks had a lot in common: extremely high leverage, failure to take advantage early
26 on to raise capital, large exposure to credit risk, and heavy reliance on funding that could quickly
disappear in a pinch. Rather than put pressure on Lehman to replenish its capital, the SEC
continued to act as if the firm had met its requirements for minimal capital and liquidity. And,
the Fed allowed Lehman to avoid the consequences of outflows by sophisticated investors by
offering to lend at below market rates through the discount window. The fact that Lehman still
could not continue as an ongoing concern for more than six months after receiving such extreme
assistance is evidence that regulators took action to help what was essentially an unsound
business. Would they have stepped in to help had Lehman management stated that the firm was
economically insolvent? Only after months of observing a clear downward trajectory did the Fed
abandon the idea that Lehman was a fundamentally sound business and allow them to enter
bankruptcy unimpeded.
Furthermore, assistance to a financial firm with a doubtful portfolio may have little
positive impact on the market as a whole. There is little that regulators can do to convince
investors that the portfolio is actually sound, which in turns means assistance in the case of
information contagion does not have many positive spillover effects. Even injecting capital
directly into the insolvent institution is ineffectual because investors are not privy to the
information that would allow them to judge that the problems have ended. Japan’s Lost Decade
is a good example of this situation: When the government of Japan injected 900 billion yen
worth of capital into Daiichi-Kangyo Bank in 1999, the firm wrote down assets by a nearly equal
amount. Investors immediately wondered if the bank would have written down losses of one
trillion yen if the government had injected another 100 billion of equity into Daiichi. Not
surprisingly, the banking woes in Japan continued for several more years after the capital
infusion. Similarly, assistance to AIG in September 2008 did not reveal the full extent of its
27 losses and its troubles continued, resulting in a new bailout with more lenient terms less than two
months later. In contrast, allowing the distressed firm to enter bankruptcy initiates a process of
evaluating the firm’s assets that would not begin outside of bankruptcy (i.e., as long as
equityholders in the financial firm focus on the option to delay and disavow insolvency).
The only comparable alternative to bankruptcy that would simultaneously eliminate the
equityholder’s incentive to obfuscate and which would remove further doubts about the quality
of the portfolio would be an assisted merger. If the merger is FDIC-assisted, this is nearly
equivalent to a financial firm bankruptcy. If the merger is assisted in a less explicit framework,
such as the New York Fed stepping in to assist LTCM or the Treasury and the Fed stepping in to
assist Bear with the JP merger, then the problem of the equityholder option to delay is eliminated
but investors learn little about the portfolio (except that is very bad) and that it may now reside
with the merger partner. Bankruptcy shines a light on these bad assets whose value would
otherwise remain extremely murky. Reducing the uncertainty surrounding the bankrupt firm’s
assets allows investors in other firms to better understand those assets and thus reach a bottom in
terms of financial distress. Without a full understanding of the losses in the portfolios, these
firms are unlikely to move forward and receive sufficient support from investors to allow them to
make more loans.
III. Past Financial Firm Failures
Intervention in a financial crisis may be warranted to avoid systemic risk, but it may also
be unproductive if it fails to stem subsequent troubles and mainly serves to send a message that
excessive risk taking will be supported by the government. The benefits of intervention are
28 difficult to gauge when we cannot know the counterfactual – what if Bear had been allowed to
file for bankruptcy? What if AIG went back to its largest shareholders for equity instead of
taking a Fed loan? Many who favor intervention point to the Lehman bankruptcy as a glaring
mistake that caused a massive downturn in financial markets and proof of the usefulness of
TBTF policy. Unfortunately, the sharp downturn in financial markets that followed Lehman’s
Chapter 11 is hard to attribute to that bankruptcy filing given that news of AIG’s troubles came
out at the same time. Lacking the counterfactual for these instances, we instead focus on past
financial firm failures to see how they affected the economy. Next we consider the cases of
Finova, Drexel, and Bank of New England (BNE).
Although quite large, Finova was not a complex financial firm. At one time a captive
finance subsidiary of Greyhound Bus, over the years Finova became a stand-alone entity that
specialized in credit to middle market firms. Finova raised capital for its business from bank
loans and public bonds. Finova’s troubles lay in its rapid growth (fivefold over a seven year
period) and, as became apparent later, relatively weak underwriting standards to a fairly
concentrated group of firms. By the time of its bankruptcy filing in March 2001, Finova’s
liabilities consisted of more than $11 billion in bonds and bank debt (its largest single debt was a
loan made by Bank of America for $295 million). Because it had neither deposits nor securities
accounts, Finova was free to file for Chapter 11 bankruptcy protection and involve all of its
business in the reorganization plan. Although technically Chapter 11 is designed as an avenue
for restructuring, in the case of financial firms the odds of leaving Chapter 11 in tact are quite
low. Not surprisingly, then, Finova’s Chapter 11 case was never a path to reorganization, but
rather a plan to liquidate. From the onset of the case, Finova shut down its underwriting
operations, and the firm’s employees were retained only for the purpose of servicing existing
29 loans and providing information for the bankruptcy. Despite the immediate drop in economywide credit from an $11 b. financial institution, no federal regulator expressed a concern for the
spillover effects of Finova’s filing nor was there any doubt that TBTF did not apply to this firm.
Its filing did not elicit concern from regulators about the impact on Bank of America, or other
creditors such as Chase, Credit Suisse and Bank One, presumably because the risk of
counterparty risk, as is typical for these diversified creditors, was minimal. Finova did not owe
much money to retail investors, so there was no discussion of a lack of confidence in the capital
markets. Its most prominent borrowers, the vacation time share firms and airlines, were already
in a slump and few would have argued that the drop in credit to these sectors would trigger a
financial accelerator effect – it was clear the firms were in no position to expand their
investments had their borrowing capacity suddenly shot up.
Finova is a textbook example of the plodding process of bankruptcy whose sluggishness
reflects the back and forth between creditors over how to maximize the size of the pie, and if not
already clear, how to divvy up that pie. Finova filed for Chapter 11 in March 2001 and exited
bankruptcy in August 2001. However, that short time period in bankruptcy was not an indication
of rapid liquidation. They simply exited Chapter 11 with a plan to liquidate all assets and pay off
creditors. The time spent in bankruptcy was mainly spent debating how to best liquidate the
firm: Some among them were convinced that the liquidation would go most successfully if it
were carried out by someone with strong bargaining power and great expertise in financial
valuation, such as Warren Buffett. Others thought those factors were important but that the main
concern was the price charged for such expertise – if the management fee ate up all the extra
increase in asset sales prices, then the creditors would be no better off. As the debate raged as to
how to dispose of the assets, none were sold. Eventually, the creditors were convinced of a
30 particular plan for liquidation and the firm exited Chapter 11. However, leaving Chapter 11 in
August 2001 was not the end of its period of distress, but merely the beginning of the end as it
slowly sold off its portfolio. As of its latest 10-K filing (recorded with the SEC in March 2009),
the company was still in the process of disposing of its assets in an orderly fashion.
Finova’s experience with liquidation under the aegis of the bankruptcy court is
instructional in light of the common perception of bankruptcy articulated by regulators. As an
example, consider Alan Greenspan’s testimony to Congress10 in defense of the Fed’s actions to
bail out LTCM. Noting the effect of the Russian default a few months previous to the collapse
of LTCM, Greenspan stated “financial markets were already unsettled by recent global events.”
Thus he believed that “had the failure of LTCM triggered the seizing up of markets, substantial
damage could have been inflicted on many market participants, including some not directly
involved with the firm, and could potentially have impaired the economies of many nations,
including our own.”11 How might LTCM’s collapse lead to systemic risk? Apparently that
mechanism hinged on excessive pressure to quickly dispose of assets in bankruptcy, as
Greenspan goes on to testify that “our sense was that the consequences of a fire sale triggered by
cross-default clauses, should LTCM fail on some of its obligations, risked a severe drying up of
market liquidity…. [and] that it was to the advantage of all parties…to engender if at all possible
an orderly resolution rather than let the firm go into disorderly fire-sale liquidation.” Greenspan
viewed the Fed-orchestrated bailout as the only way to avoid bankruptcy and thus a disorderly
liquidation. Likewise, Bernanke cited the same risk of disorder as a rationale for the Fed’s
intervention in the business of AIG:
10
Testimony to the House Committee on Banking and Financial Services, October 1, 1998, entitled “Private Sector
Refinancing of the Large Hedge Fund, Long-Term Capital Management.”
11
See Dowd (1999) for a detailed description of the LTCM bailout.
31 “In the case of AIG, the Federal Reserve, with the support of the Treasury,
provided an emergency credit line to facilitate an orderly resolution. The Federal
Reserve took this action because it judged that, in light of prevailing market
conditions and the size and composition of AIG’s obligations, a disorderly failure
of AIG would have severely threatened global financial stability, and
consequently, the performance of the U.S. economy.”12
While the Fed views the bankruptcy process as inherently disorderly, empirical evidence
suggests otherwise. Ironically, many creditors have complained that the process is too slow as
judges often look for a way to salvage assets that truly ought to be liquidated (e.g., Weiss
(1990)).
In the case of Finova, a devil’s advocate might point out that the firm did not fall into the
category of TBTF and thus its liquidation would by definition go smoothly. This is a difficult
claim to dispute, given that regulators have not precisely articulated TBTF policy. Not only do
we always face some doubt as researchers as to what constitutes “big enough” to warrant
intervention, but this doubt is intentionally engendered by regulators as a way to minimize the
impact of TBTF on moral hazard problems. Stern and Feldman (2006) argue that this problem is
best solved by convincing creditors of banks that they truly are at risk of loss – “convincing
creditors that their bank is not too big to fail (TBTF).”
Regulators seem to have taken this advice to heart in the case of Drexel, which filed for
bankruptcy in February 1990, within days of approaching regulators for assistance in hopes of a
bailout. According to Herring (2003), the problems at Drexel were mainly at the holding
company, Drexel Burnham Lambert Group (DBLG), whereas the regulated broker/dealer
subsidiaries, Drexel Burnham Lambert (DBL) and Drexel Burnham Lambert Government
Securities (DBLGS), and the commodities and foreign exchange group, Drexel Burnham
12
See testimony to Committee on Banking, Housing and Urban Affairs, U.S. Senate Sept .23, 2008. 32 Lambert Trading Corporation (DBLT), were well capitalized entities. Turmoil in the junk bond
market (due to higher defaults, S&Ls being forced to sell their portfolios, and, possibly, the
forced departure of Milken), as well as lower capital from having paid hundreds of millions in
fines to the SEC penalties, was putting downward pressure on the entire operation, particularly at
DBLG, which was having trouble continuing to fund its junk portfolio in the commercial paper
market. As an attempt to stave off distress, DBGL started upstreaming funds from its subsidiary,
DBL, claiming that the funds were excess capital. If Herring is correct that the subs were well
capitalized, then these funds were indeed excess capital. However, the SEC and other regulators
had a more pessimistic view of DBL’s capital, so when they learned of the outflows of cash from
DBL to DBGL, they viewed it as having enormous potential to cause distress to the
broker/dealer’s customers. If DBL was overstating its capital by understating the losses on its
junk bond portfolio, then DBGL was upstreaming cash that would serve as a buffer in the
liquidation by SIPC of the DBL’s accounts. The SEC stepped in and halted transfers from DBL
to DBGL, forcing DBGL to default on its debts and file for bankruptcy within a week.
Unlike Finova, many of Drexel’s assets were fairly liquid securities and quickly wound
down. In addition, many of the assets resided with the subsidiaries, which did not file for
bankruptcy. Within days, the customer accounts at DBL were transferred to Smith Barney. The
firm had been a market maker in over 200 equities, and quickly sold off its inventory, with other
market makers taking up the slack. Drexel had a derivatives book of $30 billion notional value.
Because derivatives, repos, futures and similar contracts are exempt from the bankruptcy process
(they can continue to trade, or if they are under an International Swap Dealers Association
(ISDA) contract they likely will automatically terminate upon bankruptcy), and because the
broker/dealer subs did not enter Chapter 11, these contracts were wound down outside of
33 bankruptcy over a short period of time. The Bank of England stepped in to ascertain that the
DBLT positions in currency markets were liquidated in an orderly manner and the Federal
Reserve Bank of New York helped to unwind the derivatives.13 All of the liquid securities were
converted to cash, while the illiquid junk bond portfolio remained with Drexel for more than a
year in bankruptcy.
Although Drexel was described prior to bankruptcy as a firm with more than $40 billion
in assets, in its bankruptcy filing Drexel only listed assets and liabilities of less than $4 billion
each. This reflects the fact that much of the actual business of Drexel did not enter Chapter 11
and the fact that many financial contracts, such as derivatives, are exempt from the bankruptcy
process. Compared to Finova, the more complicated, capital markets nature of the investment
bank made its bankruptcy simpler. With the desire to avoid bankruptcy for its broker/dealer
subsidiaries, which would have been instantly taken over by SIPC, and the exemptions from
bankruptcy for its more complicated financial transactions, the entity that entered Chapter 11 was
a stripped down version of Drexel. Yet, like Finova, there was no rush to sell off the junk bonds
in a fire sale and no sense of a disorderly liquidation. Items that sold easily were sold quickly,
while the less desirable assets were slowly sold off as their markets rebounded.
A third case of financial firm failure is that of the Bank of New England. As a bank
holding company whose major assets were FDIC-insured depository institutions, BNE’s case
reveals much less about how an unregulated financial institution’s bankruptcy might transmit
systemic risk. However, as a holding company, BNE was involved in operations that were
13
Interestingly, in the more recent Lehman filing, the ISDA held a four hour netting session just before the firm filed
for bankruptcy to reduce the number of contracts outstanding. This suggests the role of the Federal Reserve in
future financial firm bankruptcies will be even smaller. 34 representative of financial institutions in general. In particular, at the time of its failure, BNE
had a derivatives book with a notional value of $30 billion, mainly related to foreign exchange
rates and interest rates. According to Peek and Rosengren (1997), this placed BNE among the
top fifteen largest commercial banks in terms of foreign exchange derivatives and in the top
twenty for interest rate derivatives in the first quarter of 1990, about a year before its failure.
Over the 12 months leading up to its seizure by the FDIC in January 1991, the bank was
deteriorating due to losses on real estate. A new chairman, Lawrence Fish, was brought in to
manage the bank in February 1990, after having reported losses of more than $1 billion. In
October 1990, Institutional Investor magazine predicted that BNE would “become the next
major U.S. bank failure.” Clearly, this bank was on the brink of insolvency if not actually
insolvent in the year leading up to actual takeover by the FDIC, so one might expect that such a
large derivatives position might have sent BNE over the edge or that its demise might at least
cause massive losses on those positions. Under no circumstances could anyone describe the
FDIC seizure as sudden, indicating that the derivatives markets had many months to consider the
impact of a BNE bankruptcy on the system and to reconsider the benefits of continuing to trade
with BNE. Its failure resulted in minor losses in the derivatives market and no discernible dip in
the explosive growth of the interest rates swaps market in the early 1990s.
IV. Policy Responses
The appropriate policy response of the regulators depends in large part on the nature of
the crisis. If one firm’s distress is truly likely to bring on the collapse of a multitude of other
firms via direct business ties, then intervention to assist that one firm is appropriate. Prior
evidence suggests that financial crises are rarely of that ilk in the U.S. More often, financial
35 crises involve many firms who share common characteristics, whether it be a concentration of
assets in certain sectors of the economy such as real estate or a common reaction to high interest
rates or a shared problem with fraudulent accounting. In these cases, the appropriate policy
response would attempt to affect a broader constituency, and potentially allowing a small
fraction of the firms to fail without any assistance. Those failed firms should enter a legally
sanctioned liquidation process such as Chapter 11 that allows their assets to be sold off in an
orderly piecemeal procedure. Past experience shows us that the liquid, more desirable assets are
frequently sold off quickly to prevent them from dissipating value (especially those involving
former employees of the failed firms). In contrast, the less desirable assets are often sold off
slowly, as investors need time to evaluate their truth worth and bankruptcy participants attempt
to maximize recovery values by waiting for temporarily distressed assets to rebound. These
processes take time, making the prospect of a fire sale quite unlikely. In contrast, assisted
mergers are typically done in haste, with the result that firms like Bear are forced into the arms
of JP over the course of only a weekend, making it much more likely that the transaction will
involve an imprecise valuation based on extremely limited due diligence. Mergers like the
Bear/JP deal are examples of fire sales, not paths to avoiding them.
Although we can rule out assistance to firms based on the theory that they will cause a
domino effect, regulators may decide that the most useful course of action is to help out an entire
sector of the economy. The logic is that one sector of the economy is no longer functioning
properly and the lack of liquidity and stability in that market is hurtful to the entire economy. In
the case of commercial banks, it is easy to argue that a decline in the size of the banking sector
will reduce lending in the economy overall and that this will prevent firms from financing
positive net present value. Yet, our financial infrastructure is not so simplistic – helping out
36 large banks often hurts small banks who might otherwise have gained market share in a
downturn; helping banks might hurt insurance companies who traditionally have had more
conservative balance sheets; helping financial holding companies such as AIG might hurt more
focused firms, such as PNC. If one area of the financial sector suffers a meltdown, regulators
must be confident that its competitors are unable to step up in its place to provide the same
services. And such substitutions between sectors may involve frictions, meaning that one cannot
conclude that shrinkage over a few days will last indefinitely without regulatory intervention.
Rather, intervention, should it be warranted, ought to come in the form of broad support for all
financial intermediaries rather than for the one sector that has lost the most money.
Even assisting the financial sector as a whole involves playing favorites with financial
firms that provide external debt financing. The bias is most evident in the case of home
mortgages – firms that provided extreme amounts of home financing were bailed out in the name
of helping homeowners who could no longer afford to service this external debt. No such help
was considered for homeowners who paid cash for their houses or those whose debt servicing
was lower because they purchased a more modest house. Likewise, on the corporate side,
assistance to commercial banks in hopes that they will expand their commercial lending and
assistance to investment banks so that it might grease the wheels of the credit markets plays
favorites with firms that rely heavily on external debt financing. Further, the credit markets do
not include the venture capital market, which provides external equity financing to firms.
Assistance to either market does little for firms that traditionally rely on internally generated
equity (i.e., profits) to fund projects. Even such generic assistance to financial markets creates
the problem of moral hazard by encouraging firms to rely on external debt financing with the
hope that these debts will someday be forgiven.
37 V. Conclusion
Financial firm failures grab headlines and often generate a sensation of panic and crisis,
leading regulators such as the Fed and the Treasury to conclude they must intervene. While they
try to limit the “heads I win, tails you lose” mentality of a moral hazard problem, no regulator
wants to risk being on watch while the next Great Depression occurs. Researchers still debate
the mechanism by which the bank failures of the 1930s caused massive losses on the real side of
the economy, making it difficult for regulators to know whether their efforts to intervene are
helpful to the overall economy.
While we are still not sure of the path by which financial crises are connected to the real
economy, we can rule out the idea that financial firm failures spread to the rest of the economy
via counterparty risk. Most firms, whether financial or nonfinancial, are diversified. We rarely
see the bankruptcy of one firm causing major losses at another firm, let alone the failure of a
second, third and fourth firm. Cascades can only arise when firms’ loans to other firms are very
large as a fraction of their capital, a notion that is both at odds with bank regulations and good
business practices regarding diversification.
More likely, financial crises involving several financial firms represent commonality in
their asset allocation decisions. In these cases, regulatory aid to one firm is of little use to the
entire economy. Such assistance might bolster confidence, but clearly increasing confidence
among all such firms is more productive than merely attempting to boost confidence in one
particularly weak firm. However, confidence in the system is unlikely to be restored when
investors know that management and equity have good reason to overstate a financial firm’s
health. Admitting that the portfolio is underwater is tantamount to the CEO asking to be fired.
38 Shareholders want the option to continue as a going concern, which is nearly impossible in
bankruptcy and quite limited in an assisted merger. These players will argue that short-sellers
are destroying confidence even when they know their firms are insolvent, all in the hopes of
hanging on should their portfolios rebound. Assisting an entire sector may be rewarding poor
credit analysis and preventing more efficient lenders from appropriately gaining market share.
We must be sure that assistance to a particular sector does enough to prevent large real side
effects that we can overlook the dilatory effects of supporting the market’s least efficient
participants.
The urge to intervene in such crises is great, as regulators wish to always have stable,
liquid and orderly markets. Artificial stability, however, always comes in the form of propping
up failed firms, as regulators never intervene to pop asset bubbles in good times. Thus, stability
and liquidity achieved by crisis intervention will give the appearance of order, or at least less
disorder, as the economy recovers from the shock, but only at the expense of efficiency and
proper working order over the long haul. In the meantime, our financial firms have many
competitors who may be able to step in to replace the lost credit supply and who may provide
nearly equally desirable financial products to retail investors. Bailing out failed firms penalizes
these competitors who followed more prudent strategies before the crisis and sends a message to
failed firms that their failures do not prevent them from continuing in the business.
Regulators’ desire to maintain stable, liquid and orderly markets is best satisfied by
letting financial firms file for bankruptcy protection. The bankruptcy process allows for a much
more flexible liquidation procedure than an assisted merger, as desirable assets leave the failed
firm quickly while unpopular ones can be sold over months or, in the case of Finova, even years.
39 In contrast, an assisted merger puts pressure on markets as participants hastily settle on asset
values that make little sense, such as the decision to price Bear’s equity at $2 a share. These
mergers with their extremely limited analysis of the financial firm’s portfolio, the desire on the
part of the buyer to extract value from the regulators, and the desperately short time horizon
imposed on the seller are tantamount to fire sales. Fire sales in court-supervised bankruptcy
cases, however, are easily avoided.
Rather than stepping in to cure a particular financial firm’s problems, possibly at the
expense of other, more prudent financial competitors, policy should aim for more general
assistance to the economy. The more general the aid, the less likely it is to cause distortions in
the various financial sectors. Unfortunately, the sense of urgency that arises in a crisis means
that regulators feel compelled to come up with a solution in a matter of days, which ordinarily
excludes fiscal policy as a major weapon in the arsenal, given that Congress is not able to act as
fast as the Fed or Treasury. However, if financial regulators allow financial firms to enter
bankruptcy and slowly wind down their distressed portfolios, they could in the meantime press
Congress to implement broader-based recovery strategies.
40 References
Bebchuk, Lucian, and H.Chang (1992) “Bargaining and the division of value in corporate
reorganizations,” Journal of Law, Economics and Organization, 8, 253-279.
Benston, George J. (1986) “Federal Regulation of Banking: Historical Overview,” in George G.
Kaufman and Roger C. Kormendi, eds., Deregulating Financial Services: Public Policy in
Flux (Ballinger Press, Cambridge, MA).
Bernanke, Ben S. (1983) “Nonmonetary Effects of the Financial Crisis in the Propagation of the
Great Depression,” American Economic Review, 73, 257-276.
Bernanke, Ben S., Mark Gertler, and Simon Gilchrist (1999) “The Financial Accelerator in a
Quantitative Business Cycle Framework,” Handbooks in Economics, vol. 15.
Amsterdam: Elsevier, pp. 1341-93.
Bradley, Michael and Michael Rosenzweig (1992) “The Untenable Case for Chapter 11,” Yale
Law Journal, 101, 1043-1095.
Campbell, John Y., Martin Lettau, Burton G. Malkiel and Yexiao Xu (2001) “Have Individual
Stocks Become More Volatile? An Empirical Exploration of Idiosyncratic Risk,” ,"
Journal of Finance 61, 1-43.
Collin-Dufresne, Pierre, Robert S. Goldstein, and Jean Helwege (2004) “Is Event Risk Priced?
Modeling Contagion via the Updating of Beliefs,” working paper, Columbia University.
Covrig, Vicentiu, Buen Sin Low and Michael Melvin (2004) “A Yen is Not a Yen:
TIBOR/LIBOR and the Determinants of the Japan Premium,” Journal of Financial and
Quantitative Analysis, 39, 193-208.
Crabbe, Leland E. (1991) “Event Risk: An Analysis of Losses to Bondholders and “Super
Poison Put” Bond Covenants,” Journal of Finance, 46, 689-706.
Davis, M., and V. Lo (1999) “Infectious Defaults,” working paper, Tokyo Mitsubishi
International PLC
DeAngelo, Harry, Linda DeAngelo, and Stuart C. Gilson (1994) "The Collapse of First
Executive Corporation: Junk Bonds, Adverse Publicity, and the Run on the Bank
Phenomenon," Journal of Financial Economics 36, 287-336.
DeYoung, Robert and Jack Reidhill (2008) “A Theory of Bank Resolution: Political Economics
and Technological Change,” University of Kansas working paper.
Dowd, Kevin (1999) “Too Big to Fail? Long-Term Capital Management and the Federal
Reserve,” briefing paper #52, Cato Institute.
41 Duffie, Darrell and David Lando (2001) “Term Structures of Credit Spreads with Incomplete
Accounting Information,” Econometrica, 69, 633-664.
Eichengreen, Barry and Ashoka Mody (2000) “What Explains the Changing Spreads on
Emerging Market Debt?” in Sebastian Edwards (ed.), The Economics of International
Capital Flows (University of Chicago Press).
Fenn, George W. and Rebel A. Cole (1994), “Announcements of Asset-Quality Problems and
Contagion Effects in the Life Insurance Industry,” Journal of Financial Economics, 35,
181-198.
Flannery, Mark J. and Sorin M. Sorescu (1996) “Evidence of Bank Market Discipline in
Subordinated Debenture Yields: 1983-1991,” Journal of Finance, 51, 1347-1377.
Friedman, Milton and Anna Schwartz (1971), A Monetary History of the United States 18641960 (University of Chicago Press, Chicago, Illinois).
Furfine, Craig H. (2003), “Interbank Exposures: Quantifying the Risk of Contagion,” Journal of
Money, Credit and Banking, 35, 111-128.
Geske, Robert (1977), “The Valuation of Corporate Liabilities as Compound Options,” Journal
of Financial and Quantitative Analysis, 12, 541-552.
Giesecke, Kay (2002) “Correlated Default with Incomplete Information,” working paper,
Humboldt University
Gilson, Stuart (1989) “Management Turnover and Financial Distress,” Journal of Financial
Economics, 25, 241-262.
Haugen, Robert A. and Lemma W.Senbet (1978) “The Insignificance of Bankruptcy Costs to the
Theory of Optimal Capital Structure,” Journal of Finance, 33, 383-393.
Helwege, Jean (1996), “Determinants of S&L Failure Rates: Estimates of a Time-Varying
Proportional Hazard Function,” Journal of Financial Services Research, 7, 55-61.
Helwege, Jean and Paul Kleiman (1997), “Understanding Aggregate Default rates of High-Yield
Bonds,” Journal of Fixed Income, 7, 55-61.
Hertzel, Michael G., Zhi Li, Micah S. Officer and Kimberly J. Rodgers (2008), “Interfirm
Linkages and the Wealth Effects of Financial Distress along the Supply Chain,” Journal
of Financial Economics, 87, 374-387.
Jarrow, Robert A. and Fan Yu (2001) “Counterparty Risk and the Pricing of Defaultable
Securities,” Journal of Finance, 61, 1765-1799.
42 Jorion, Phillippe and Gaiyan Zhang (2007) “Good and Bad Credit Contagion: Evidence from
Credit Default Swaps,” Journal of Financial Economics, 84, 860-883.
Jorion, Phillippe and Gaiyan Zhang (2008) “Credit Contagion from Counterparty Risk,” UC
Irvine working paper.
Kabir, M. Humayun and M. Kabir Hassan (2005) “The Near-collapse of LTCM, US Financial
Stock Returns, and the Fed,” Journal of Banking and Finance, 29, 441-460.
Kamin, Steven B. and Karsten Von Kleist (1999) “The Evolution and Determination of
Emerging Market Credit Spreads in the 1990s“ Federal Reserve Board working paper.
Lang, Larry H.P. and Rene M. Stulz (1992) “Contagion and Competitive Intra-industry Effects
of Bankruptcy Announcements,” Journal of Financial Economics, 32, 45-60.
Peek, Joe and Eric S. Rosengren (1997) “Derivatives Activity at Troubled Banks,” Journal of
Financial Services Research, 12, 287-302.
Remolona, Eli, William Bassett, and In Sun Geoum (1996)“Interbank Lending and Systemic
Risk,” Federal Reserve Bank of New York Economic Policy Review, April, 17-38.
Rochet, Jean-Charles and Jean Tirole (1996) “Interbank Lending and Systemic Risk,” Journal of
Money, Credit and Banking, 28, 733-762.
Shleifer, Andrei and Robert Vishny (1992) “Liquidation Values and Debt Capacity: A Market
Equilibrium Approach,” Journal of Finance, 47, 1343-1366.
Smith David C. and Per Strömberg (2004) “Maximizing the Value of Distressed Asset Sales:
Bankruptcy Law and the Efficient Reorganization of Firms,” in Systemic Financial
Distress: Containment and Resolution, Patrick Holohan and Luc Laeven editors,
Cambridge University Press.
Stern, Gary H. and Ron J. Feldman (2006) “Managing Too Big to Fail by Reducing Systemic
Risk: Some Recent Developments,” The Region, Federal Reserve Bank of Minneapolis
Theocharides, George (2008) “Contagion: Evidence from the Bond Market,” working paper,
SKK Graduate School of Business, Seoul, Korea
Weiss, Lawrence A. (1990) “Bankruptcy resolution: Direct costs and violation of priority of
claims,” Journal of Financial Economics, 27, 285-314.
Weiss, Lawrence A. and Karen H. Wruck (1998) “Information Problems, Conflicts of Interest
and Asset Stripping: Chapter 11’s Failure in the Case of Eastern Airlines,” Journal of
Financial Economics, 48, 55-97.
43 Wruck, Karen H. (1990) “Financial distress, reorganization, and organizational efficiency,”
Journal of Financial Economics, 27, 419-444.
Yu, Fan (2005) “Accounting Transparency and the Term Structure of Credit Spreads,” Journal
of Financial Economics, 75, 53-84.
44